Lecture notes: QM 04

Expectation values of operators

Dr. Mohammad A Rashid

September 27, 2020
just.edu.bd/t/rashid

Contents

1 Momentum space 1

2 Expectation values of operators 4

3 Uncertainty 6

4 Time dependence of expectation values 7

References 8

1 Momentum space

We begin with Fourier transformation to make a series of developments that lead to the
idea of momentum space as a counterpoint or dual of position space. In this section the
time dependence of wave functions will play no role. Therefore we will simply suppress
time dependence. You can imagine all wave functions evaluated at time equal zero or at
some arbitrary time t0 .

A wave packet is a superposition of plane waves eikx with various wavelengths. According
to Fourier’s theorem such a wave packet have the form
Z ∞
1
Ψ(x) = √ Φ(k)eikx dk. (1)
2π −∞

The function Φ(k) acts as the weight with which we add the plane waves with momentum
}k to form Ψ(x). Thus the Fourier representation of the wave Ψ(x) gives the way to
represent the wave as a superposition of plane waves of different momenta. If we know
Ψ(x) then Φ(k) is calculable. In fact, by the Fourier inversion theorem, the function Φ(k)
is the Fourier transform of Ψ(x), so we can write
Z ∞
1
Φ(k) = √ Ψ(x)e−ikx dx. (2)
2π −∞

1
Note the symmetry in the two equations above. The Fourier transform Φ(k) has all
the information carried by the wave function Ψ(x). This is clear because knowing Φ(k)
means knowing Ψ(x).

The consistency of the above equations can be used to derive an integral representation
for a delta function. The idea is to replace Φ(k) in (1) by the value given in (2). In
order to keep the notation clear, we must use x0 as a dummy variable of integration in
the second equation. We have
Z ∞ Z ∞
1 1 0
Ψ(x) = √ dk e √ikx
dx0 e−ikx Ψ(x0 )
2π 2π
Z ∞ −∞ Z ∞ −∞
1 0
= dx0 Ψ(x0 ) dk eik(x−x ) . (3)
−∞ 2π −∞
| {z }
The factor indicated by the brace happens to reduce the x0 integral to an evaluation at
x. We know that δ(x0 − x) is the function such that for general f (x)
Z ∞
dx0 f (x0 ) δ(x0 − x) = f (x) (4)
−∞

and so we conclude that the factor indicated by the brace is a delta function
Z ∞
0 1 0
δ(x − x) = dk eik(x−x ) . (5)
2π −∞
R
In this integral one can let k → −k and since dk is left invariant under this replacement,
we find that δ(x0 − x) = δ(x − x0 ), or more plainly δ(x) = δ(−x). We will record the
integral representation of the delta function using the other sign:
Z ∞
0 1 0
δ(x − x ) = dk eik(x−x ) . (6)
2π −∞

Another useful property of delta functions is

1
δ(ax) = δ(x). (7)
|a|

To understand how the normalization condition for Ψ(x) look like in terms of Φ(k), we
calculate
Z ∞ Z ∞ Z ∞ Z ∞
∗ 1 ∗ −ikx 1 0
Ψ (x)Ψ(x) dx = dx √ Φ (k)e dk √ Φ(k 0 )eik x dk 0 . (8)
−∞ −∞ 2π −∞ 2π −∞
Rearranging the integrals to do the x integral first we write
Z ∞ Z ∞ Z ∞ Z ∞
∗ 0 ∗ 0 1 0
Ψ (x)Ψ(x) dx = dk dk Φ (k)Φ(k ) ei(k −k)x dx
−∞ 2π −∞
Z−∞
∞ Z−∞
∞
= dk dk 0 Φ∗ (k)Φ(k 0 ) δ(k 0 − k)
Z−∞
∞
−∞

= dk Φ∗ (k)Φ(k 0 ), (9)
−∞

2
where we recognized the presence of a delta function and we did the integral over k 0 . Our
final result is therefore
Z ∞ Z ∞
2
|Ψ(x)| dx = |Φ(k)|2 dk. (10)
−∞ −∞

This is known as Parseval’s theorem, or more generally, Plancherel’s theorem. This

equation relates the Ψ(x) normalization to a rather analogous normalization for Φ(k).
This is a hint that just like for |Ψ(x)|2 , we may have a probability interpretation for
|Φ(k)|2 .

Since physically we associate our plane waves with eigenstates of momentum, let us
rewrite Parseval’s theorem using momentum p = }k. Instead of integrals over k we will
have integrals over p. With Φ(p)
e = Φ(k) equations (1) and (2) can be rewritten as
Z ∞
1 e eipx/h dp,
Ψ(x) = √ Φ(p)
2π } −∞
Z ∞ (11)
1
Φ(p)
e =√ Ψ(x) e−ipx/h dx.
2π −∞

For a more symmetric

√ pair of equations we can redefine the function Φ(p).
e We let
Φ(p)
e → Φ(p) } in (11) to obtain our final form for Fourier’s relations in terms of
momentum:
Z ∞
1
Ψ(x) = √ Φ(p) eipx/h dp,
2π} −∞
Z ∞ (12)
1
Φ(p) = √ Ψ(x) e−ipx/h dx.
2π} −∞

Similarly, Parseval’s theorem (10) becomes

Z ∞ Z ∞
2
|Ψ(x)| dx = |Φ(p)|2 dp. (13)
−∞ −∞

Our interpretation of the top equation in (12) is that Φ(p) denotes the weight with
which we add the momentum state eipx/} in the superposition that represents Ψ(x). This
momentum state eipx/} is an eigenstate of the momentum operator p̂ with eigenvalue p.
Just like we say that Ψ(x) is the wave function in position space x, we can think of Φ(p)
as the wave function in momentum space p. The Parseval identity (13) suggests that
Φ(p) has a probabilistic interpretation as well. Given that a properly normalized Ψ(x)
leads to a Φ(p) that satisfies
Z ∞
|Φ(p)|2 dp = 1, (14)
−∞

we postulate that:

3
 |Φ(p)|2 dp is the probability to find the particle
with momentum in the range [p, p + dp].

This makes the analogy between position and momentum space quite complete.

Fourier’s theorem in momentum space language for three dimension takes the form
Z ∞
1
Ψ(r) = Φ(p) eip·r/h dp,
(2π})3/2 −∞
(15)
Z ∞
1
Φ(p) = Ψ(r) e−ip·r/h dr.
(2π})3/2 −∞

Just like one dimensional case using the definition of three dimensional delta function
Z ∞
3 0 1 0
δ (r − r ) = 3
eik·(r−r ) (16)
(2π) −∞

one can arrive to Parseval’s identity:

Z ∞ Z ∞
2
|Ψ(r)| dr = |Φ(p)|2 dp, (17)
−∞ −∞

where |Φ(p)|2 dp is the probability to find the particle with momentum in the range dp
around p.

2 Expectation values of operators

Consider a random variable ω that takes values in the set {Ω1 , . . . , Ωn } with respective
probabilities {p1 , . . . , pn }. The expectation value hΩi (or expected value) of Ω is the
average value that we expect to find after repeated observation of Ω, and is given by the
formula
n
X
hΩi = p i Ωi . (18)
i=1

In quantum system the probability for a particle to be found in [x, x + dx] at time t is
given by Ψ∗ (x, t)Ψ(x, t) dx. Thus, the expectation value of x, denoted as hxi is given by
Z ∞
hxi ≡ x Ψ∗ (x, t)Ψ(x, t) dx. (19)
−∞

The physical meaning of this would be if we consider many copies of the identical system,
and measure the position x at a time t in all of them, then the average value recorded will
converge to hxi as the number of systems and measurements approaches infinity. Note
that the expectation depends on time t.

4
Let us now discuss the expectation value for the momentum. Since Φ∗ (p, t)Φ(p, t) dp is
the probability of finding the particle with momentum in the range [p, p + dp] at time t,
we define the expectation value of the momentum as
Z ∞
hpi ≡ p Φ∗ (p, t)Φ(p, t) dp. (20)
−∞

We will now manipulate this expression to see what form it takes in coordinate space.
Using (12) and its complex conjugate version we have
Z ∞
hpi = p Φ∗ (p, t)Φ(p, t) dp
Z−∞∞ Z ∞ Z ∞
dx ∗ dx0 0
= dp p √ Ψ (x, t) eipx/}
√ Ψ(x0 , t) e−ipx /}
−∞ 2π} 2π}
Z ∞ −∞ Z ∞ −∞
Z ∞
1 0
= dx Ψ∗ (x, t) dx0 Ψ(x0 , t) dp p eipx/} e−ipx /}
2π} −∞
Z ∞ Z−∞∞ Z−∞
∞  
1 ∗ 0 0 } ∂ 0
= dx Ψ (x, t) dx Ψ(x , t) dp eipx/} e−ipx /}
2π} i ∂x
Z ∞ −∞ Z ∞ −∞  −∞
 Z ∞
} ∂ 1 0
= dx Ψ∗ (x, t) dx0 Ψ(x0 , t) dp eipx/} e−ipx /} . (21)
−∞ −∞ i ∂x 2π} −∞

Using p = }u in the final integral we have

Z ∞ Z ∞
1 ipx/} −ipx0 /} 1 0
dp e e = du eiu(x−x ) = δ(x − x0 ). (22)
2π} −∞ 2π −∞
As a result, we have
Z ∞ Z ∞  
∗ 0 0 } ∂
hpi = dx Ψ (x, t) dx Ψ(x , t) δ(x − x0 )
i ∂x
Z−∞
∞ −∞  Z ∞
} ∂
= dx Ψ∗ (x, t) dx0 Ψ(x0 , t)δ(x − x0 )
i ∂x
Z−∞
∞   −∞
} ∂
= dx Ψ∗ (x, t) Ψ(x, t). (23)
−∞ i ∂x
∂ } ∂
Finally with p̂ = −i} = we have shown that
∂x i ∂x
Z ∞
hpi = Ψ∗ (x, t) p̂Ψ(x, t) dx. (24)
−∞

Notice the position of the p̂ operator: it acts on Ψ(x, t). This motivates the following
definition for the expectation value hΩi of any operator Ω̂:
Z ∞
hΩi = Ψ∗ (x, t) Ω̂Ψ(x, t) dx. (25)
−∞

Example: Consider the kinetic energy operator T̂ for a particle moving in 1D:
p̂2 }2 ∂ 2
T̂ = =− . (26)
2m 2m ∂x2
5
According to the definition (25) we have
Z ∞
hT i = dx Ψ∗ (x, t) T̂ Ψ(x, t)
−∞
Z ∞
}2 ∂2
= − dx Ψ∗ (x, t) 2 Ψ(x, t). (27)
2m −∞ ∂x

The kinetic energy is a positive operator (being proportional to the square of the momen-
tum operator). It is therefore of interest to make this positivity manifest. Integrating by
parts one of the x derivatives and ignoring boundary terms that are presumed to vanish,
we find
 ∂Ψ(x, t) 2
Z ∞
}2

hT i =   dx. (28)
2m −∞  ∂x 

This is manifestly positive! The expectation value of T can also be computed in momen-
tum space using the probabilistic interpretation that led to (20):
Z ∞ 2
p
hT i = |Φ(p, t)|2 dp. (29)
−∞ 2m

3 Uncertainty

For random variables, the uncertainty is the standard deviation i.e. the square root of
the expected value of the square of deviations from the average value. Let Ω be a random
variable that takes on values {Ω1 , . . . , Ωn } with probabilities {p1 , . . . , pn }, respectively.
The expectation value is
X
Ω= pi Ωi , (30)
i

and the variance (the square of the standard deviation) is

X
(∆Ω)2 ≡ pi (Ωi − ∆Ω)2 . (31)
i

This definition makes it clear that if ∆Ω = 0, then the random variable is constant: each
term in the above sum must vanish, making Ωi = ∆Ω, for all i. We find another useful
expression by expanding the above definition
X X X 2
(∆Ω)2 = pi Ω2i − p i Ωi Ω + pi Ω
i i i
2
= Ω2 − 2Ω Ω + Ω
2
= Ω2 − Ω , (32)

pi = 1. Since, by definition (∆Ω)2 ≥ 0, we have interesting inequality

P
where we used i

2
Ω2 ≥ Ω . (33)

6
Now let us consider the quantum mechanical case. Following (32) we declare that the
square of the uncertainty ∆Ω of a variable is a real number whose square is given by
(∆Ω)2 = hΩ2 i − hΩi2 . (34)
Hence the uncertainty is given as
p
∆Ω = hΩ2 i − hΩi2 . (35)

Since the right-hand side of (34) can be written as

hΩ2 i − hΩi2 = hΩ2 i − 2hΩihΩi + hΩi2 = hΩ2 − 2ΩhΩi + hΩi2 i, (36)
the square of uncertainty can also be written as the expectation value of the square of
the difference between the operator and its expectation value:
(∆Ω)2 = h(Ω − hΩi)2 i. (37)

4 Time dependence of expectation values

The expectation values of operators are in general time dependent because the wave
functions representing the states are time dependent. We will consider here operators
that do not have explicit time dependence, that is, operators that do not involve any
time derivatives. Taking the time derivative on both sides of (25) and then multiplying
with i} we have
d ∞ ∗
Z
d
i} hΩi = i} Ψ (x, t) Ω̂Ψ(x, t) dx
dt dt −∞
Z ∞ ∗ 
∂Ψ ∗ ∂Ψ
= i} Ω̂Ψ + Ψ Ω̂ dx. (38)
−∞ ∂t ∂t
∂Ψ
Now using the Schrödinger equation i} = ĤΨ and its complete conjugate we have
∂t
Z ∞n
d o
i} hΩi = −(ĤΨ)∗ Ω̂Ψ + Ψ∗ Ω̂(ĤΨ) dx
dt
Z−∞
∞ n o
= Ψ∗ Ω̂(ĤΨ) − (ĤΨ)∗ Ω̂Ψ dx. (39)
−∞

Recall the Hermiticity of Ĥ, which implies that

Z ∞ Z ∞
∗
(ĤΨ1 ) Ψ2 dx = Ψ∗1 ĤΨ2 dx. (40)
−∞ −∞

This can be applied to the second term in the last right-hand side of (39) to move Ĥ
into the other wave function
Z ∞
d
i} hΩi = (Ψ∗ Ω̂ĤΨ − Ψ∗ Ĥ Ω̂Ψ) dx
dt
Z−∞
∞
= Ψ∗ (Ω̂Ĥ − Ĥ Ω̂)Ψ dx
Z−∞
∞
= Ψ∗ [Ω̂, Ĥ]Ψ dx, (41)
−∞

7
where we noted the appearance of the commutator. All in all, we have proven that for
operators Ω̂ that do not explicitly depend on time,

d D E
i} hΩi = [Ω̂, Ĥ] . (42)
dt

Therefore, for an operation which commute with the Hamiltonian operator Ĥ the expec-
tation value will not change over time. For example for a free particle the Hamiltonian
operator is

p̂2
Ĥ = . (43)
2m

Hence the momentum operator p̂ commute with Ĥ that is

p̂2 1 2 1  
[p̂, Ĥ] = [p̂, ]= [p̂, p̂ ] = [p̂, p̂]p̂ + p̂[p̂, p̂] = 0. (44)
2m 2m 2m
This gives us
d D E
i} hpi = [p̂, Ĥ] = 0. (45)
dt
For a free particle the expectation value of momentum does not change over time i.e. for
a free particle the the expectation value of momentum is conserved.

Note: Most of the materials in this lecture note are taken from the lecture on Quantum
Physics by Prof. Barton Zwiebach for the course 8.04 in the year of 2016 at MIT, USA.

References
1. Quantum Mechanics by Nouredine Zettili
2. Introduction to Quantum Mechanics by David J. Griffiths